Interfacial doping of carbon nanotubes at the polarisable organic/water interface: a liquid/liquid pseudo-capacitor

The electrochemical reactivity of single-walled carbon nanotube (SWCNT) films, assembled at a polarisable organic/water interface, has been probed using model redox species. Electrons generated by the oxidation of organic 1,10-dimethylferrocene (DMFc) to DMFc can be transferred through the assembled SWCNT layer and reduce aqueous ferricyanide (Fe(CN)6 3 ) to ferrocyanide (Fe(CN)6 4 ), with a doping interaction observed. Several electrochemical techniques, including cyclic voltammetry and electrochemical impedance spectroscopy (EIS), were employed to confirm that the model redox couples dope/charge the SWCNTs. In situ Raman spectro-electrochemistry was also applied to verify the charge transfer processes occurring at the assembled SWCNT films and confirm that the doping effect of the carbon nanotubes is initiated by electrochemical reactions. This doping interaction indicated that the adsorbed SWCNT films can act as a pseudo-capacitor, showing a high area-normalised capacitance. The deeper understanding of the electrochemical properties of SWCNTs, gained from this study, will help determine the performance of this material for practical applications.


Introduction
Over the last few decades, 1 carbon nanotubes (CNTs) have received considerable attention from the scientic community due to their varying and fascinating range of properties.These materials are composed of sp 2 -hybridised carbon sheets (graphene) wrapped into a tube structure, with tube diameters on the order of a few nm's. 2,3A CNT formed from a single wrapped graphene sheet is termed a single-walled CNT (SWCNT), while those formed of multiple layered graphene sheets are termed multi-walled CNTs (MWCNTs). 4Graphene sheets can wrap in various orientations to form CNTs, leading to differing helicities, which inuence the electronic density of states within the CNTs and, in turn, lead to either a metallic or semiconducting electronic band structure. 2,4The properties of CNTs are also affected by the open, or closed, nature of the tube ends.In addition to their variety, the mechanical and electronic properties of CNTs are individually impressive, with high tensile strength and Young's modulus, a surface area up to 1500 m 2 g À1 , thermal conductivity of 6000 W m À1 K À1 and, in metallic CNTs, ballistic electron transfer. 3,5To fully utilise these exceptional and changeable properties, the fundamental study of CNTs is of great importance.
Many applications, for which CNTs may be suitable, require assembly of the nanotubes into larger coherent structures.Liquid/liquid systems, composed of immiscible aqueous and organic phases of differing density, are ideally suited for this purpose.2][13][14][15][16][17] The versatile chemistry of liquid/liquid systems has also been cleverly utilized by the group of Zarbin to form composite structures such as CNT/polyaniline lms. 18,19issolution of supporting electrolyte in either liquid phase allows liquid/liquid systems to be electried.This particular form of liquid/liquid interface is known as the interface between two immiscible electrolyte solutions (ITIES).Application of a Galvani potential difference across the ITIES can then be used to drive reactions between the two phases.Such systems are analogous to the more commonly encountered solid electrode/liquid systems, with current generated by the passage of charge across the ITIES.][22][23] Building on the work on liquid/liquid assembled CNTs, some recent studies have utilized ITIES systems to study the electrochemical properties of CNTs and other similar carbon materials.5][26] Various interfacially adsorbed materials have been used as catalysts for these reactions and the group of Girault found that using MWCNTs 27 and reduced graphene oxide (rGO) 28 as supports for the catalysts Mo 2 C and MoS 2 , respectively, improved the catalytic effect of these materials, compared to their unsupported forms, toward the HER.These effects were attributed to an improved electron transfer from the lipophilic reducing agent to the interfacial catalyst, mediated by the carbon supports.Interfacial graphene 29 and graphene oxide (GO) 30 have also been found to display inherent catalytic properties toward the ORR at the ITIES.An additional promising, and recently found, candidate for the study of interfacial charge transfer processes is the water/room temperature ionic liquid interface. 31,32Such interfaces can have an impact in areas such as proton-coupled electron transfer, fuel cells, and hydrogen storage, where ionic liquids are used as aprotic solvents. 335][36] In combination with an ex situ metal deposition step, prior to assembly at the ITIES, this methodology was used to asymmetrically decorate both sides of a chemical vapour deposition (CVD) grown graphene sheet with Au and Pd NPs. 37An interesting consequence of inserting CVD graphene to the ITIES is the proposed suppression of capillary waves, which exist at the bare ITIES and cause roughening of the interface.This phenomenon was suggested by an observed decrease in the interfacial capacitance in the presence of graphene, as measured by electrochemical impedance spectroscopy (EIS). 36Additionally, the growth of a CNT/polypyrrole composite lm was controlled electrochemically at the ITIES. 38aman spectroscopy is one of the most widely used techniques for characterisation of CNTs, with the spectra providing a plethora of information on the electronic structure and diameter of the tubes and their defect density. 39Several bands are present in the Raman spectra of CNTs, with the key bands being the radial breathing mode (RBM), the tangential displacement mode (G), the high-frequency two-phonon mode (2D or G 0 ) band and the defect-induced D band.1][42][43][44] Very recently, Ibañez et al. 45 reported the use of Raman spectro-electrochemistry to study charge transfer processes at the ITIES, leading to the possibility of studying the state of CNTs and graphene during mediation of heterogeneous electron transfer at the ITIES; as the interfacial ET initiated charge transfer will result in a change in the Fermi-level of the assembled carbon nanostructures, thus doping them.All of these phenomena are of interest in electro-catalytic reactions, proton-coupled electron transfers and energy storage using the nanostructured organic/water interface.
In the current work we present a study of the interfacial charging interaction between model redox species and SWCNTs assembled at the ITIES, using several electrochemical techniques and in situ Raman spectro-electrochemistry.Additionally, the interfacial SWCNT density, and thus the SWCNT lm thickness, was varied.

Methods
Electrochemical experiments were performed using a fourelectrode conguration with a PGSTAT20 (Eco Chemie Autolab) Scheme 1 ITIES cell compositions for the supporting blank (A), both aqueous and organic redox couples (B) and only the aqueous redox couple (C) configurations and respective schematics (D-F) when the CNT layers are assembled at the ITIES (DCB/water).
Ag/AgCl reference electrodes (REs) were directly immersed in the chloride-containing aqueous phase, while an aqueous solution of 0.1 M LiCl and 1 mM BTPPACl was brought into contact with the organic solution to form a second liquid/liquid junction for the organic reference electrode.The three-arm cell used for the liquid/liquid electrochemical measurements, had an interfacial area of 0.74 cm 2 and a total solution volume of 2.5 mL.The Galvani potential scale (D w o F) on all cyclic voltammograms (CVs) was normalized using the formal transfer potential of TMA + , which corresponds to D w o F ¼ 0.277 V. 48 A three-electrode conguration was employed as a control to determine the diffusion coefficient for DMFc oxidation, using a platinum microdisc electrode (25 micron in diameter; CH Instruments Inc.) as the working electrode and a standard Pt mesh electrode and a Ag wire as the counter and reference electrodes, respectively.
SWCNT powder (2 mg mL À1 ) was dispersed in DCB using ultrasonication in a bath sonicator (Elmasonic P70H) operating at 37 kHz and 40% power for 12 h.A recirculating cooling system (Julabo F250) was used to maintain a stable temperature of 20 C throughout sonication.UV-Vis spectroscopy (DH-2000-BAL, Ocean Optics and USB2000 interface, Micropack GmbH) was used to measure the concentration of the SWCNT dispersions.
The assembly of SWCNTs at the organic/water interface in a three-arm cell is described elsewhere. 34,35Briey, an aliquot of SWCNT dispersion, in a DCB solution containing the organic electrolyte, was placed in contact with an equal volume of aqueous phase, and self-assembly of the SWCNT at the interface was then achieved using a short (less than 5 min) sonication.In situ Raman spectro-electrochemical measurements were carried out using the method reported by Ibañez et al. 45 To summarise, Raman spectra were collected using a Confocal Raman Voyage (BWTEK) using a 20Â objective with a laser wavelength of 532 nm (operating power of 15 mW).An XYZ piezoelectric positioner (Newport 271), controlled by a Newport motion controller (Newport, ESP 301), was employed to focus the laser beam with micrometric resolution.The organic phase was gellied using high molecular weight polyvinylchloride (PVC), following the methodology reported previously. 49

Electron transfer at interface assembled SWCNT lms
The effect of adsorbed SWCNT lms on biphasic electron transfer, with one redox reaction in the aqueous phase (aq) and the opposite redox reaction in the organic phase (o), at the ITIES is probed here.Well-known model redox systems were used to study the effect of assembled SWCNT lms on such redox processes; explicitly, the reduction of ferricyanide (Fe(CN) 6 3À ) to ferrocyanide (Fe(CN) 6 4À ) and the oxidation of DMFc to DMFc + in the aqueous and organic phases, 50 respectively, as stated in reaction (1).
SWCNTs were dispersed in the organic phase at a range of concentrations and, aer assembly at the ITIES, produced interfacial SWCNT lms with densities of 1.38, 2.77, 5.55, 11.1, and 22.2 mg cm À2 , as calculated using the method reported by Rodgers et al. 29 The SWCNT lm completely covers the interface at the lowest concentration used here (1.38 mg cm À2 ), so increasing the density of SWCNTs will likely increase the thickness of the interfacial SWCNT lm.
Voltammetric measurements were carried out with 0.1 M LiCl and 10 mM BTPPATPBCl supporting electrolytes employed in the aqueous and organic phases, respectively (Fig. 1).A blank CV, in the absence of SWCNTs at the ITIES (Scheme 1A and D), shows a potential window limited by transfer of the aqueous supporting electrolyte ions, Li + and Cl À (Fig. 1a).
Interfacial electron transfer (ET) occurred when both redox couples were present (Fig. 1b), represented by the faradaic peaks between 0.0 V and +0.1 V, which correspond to heterogeneous DMFc oxidation and re-reduction of DMFc + (Scheme 1B).A diffusion coefficient for the oxidation (D ox ) of DMFc to DMFc + at the DCB/water interface, was calculated using the Randles-Sevčik equation (eqn (2)): 51,52 I p ¼ 0:4463 where I p is the peak current, R is the gas constant, T is the temperature, z is the charge, F is Faraday's constant, A is the area of the interface, c is the concentration of the limiting reagent and n is the scan rate.A value of D ox ¼ 4.05 Â 10 À6 cm 2 s À1 was found in the absence of assembled SWCNTs, which is somewhat lower than the literature value of D ox ¼ 8.08-9.64Â 10 À6 cm 2 s À1 . 34,53Additionally, the diffusion coefficient of 500 mM DMFc undergoing oxidation solely in DCB (containing 10 This journal is © The Royal Society of Chemistry 2016 mM BTPPATPBCl supporting electrolyte) was independently determined using microelectrode voltammetry according to eqn (3), 51 where r is the radius of disc (r ¼ 12.5 mm) and i is the diffusion limited current, and found to be 7.09 Â 10 À6 cm 2 s À1 .This value is similar to previous works which reported D ox ¼ 7.17 Â 10 À6 cm 2 s À1 . 54¼ 4zFDcr (3) Fig. 1c-g shows reversible CVs obtained in the presence of different coverages/densities of SWCNTs at the interface (b: ET blank ¼ 0 mg cm À2 , c: 1.38 mg cm À2 , d: 2.77 mg cm À2 , e: 5.55 mg cm À2 , f: 11.1 mg cm À2 , g: 22.2 mg cm À2 ).The current increases with increasing SWCNT surface coverage/thickness, which may be the result of a doping effect on the SWCNTs, or a result of a change in surface area at the interface. 34Additionally, a small shi can be seen in the peak position, which could also come from doping of the SWCNTs, due to the redox reaction.
In order to probe the charging effect of the interfacial ET process on the SWCNT lms, the voltammetric analysis was repeated in the absence of the organic DMFc redox species (Scheme 1C and F).Fig. 2A(b-f) shows cyclic voltammetric characterisation of the cells containing Fe(CN) 6 3À /Fe(CN) 6

4À
, in the presence of SWCNT lms of differing density/thickness.The current increases as a function of the SWCNT concentration in the presence of the aqueous phase redox couple, suggesting that the potential difference applied across the ITIES causes the reduction of Fe(CN) 6 3À to Fe(CN) 6 4À via the interfacial SWCNTs, thus resulting in electrochemical doping of the SWCNTs. 55An alternative explanation is that the current increase is due to a growth in capacitive current, resulting from the increase in the SWCNT lm thickness.
In order to further probe the mechanism of the interfacial charging, several cyclic voltammetric and electrochemical impedance spectroscopy (EIS) measurements were performed in the presence of interfacial assembled SWCNT lms of varying density/thickness.The electro-activity of the SWCNT lms, with only the supporting electrolytes in both phases (Scheme 1A and D), is shown in Fig. 2B.The similarity in capacitive current, at each different SWCNT lm thickness, and the absence of any faradaic peaks (Fig. 2B(b-f)), demonstrates that capacitive charging of the interfacial SWCNTs is not responsible for the current increase seen in Fig. 2A.
EIS measurements were performed in the three different DCB/water congurations: only the supporting electrolytes (Scheme 1A and D), redox species in both phases (Scheme 1B and E), and only the ferri/ferrocyanide redox couple (Scheme 1C and F); and at varying potential differences, corresponding to the completely reduced (À0.1 V), reduced (+0.05V), oxidised (+0.15 V), and completely oxidised (+0.35 V) forms of DMFc.The representative complex impedance Nyquist plots at +0.15 V and +0.05 V, corresponding to heterogeneous DMFc oxidation and re-reduction of DMFc + , respectively (Scheme 1B), proceeding via a 5.55 mg cm À2 SWCNT lm at the DCB/water interface are shown in Fig. 3.The impedance data were analysed and tted using the equivalent electrical circuit 56,57 which is depicted by Fig. 3, where C int is the capacitance of the interface, R CT is the  charge transfer resistance, Z W is the Warburg impedance, C st is the stray capacitance and R s is the solution resistance.
Calculated C int values are plotted vs. the interfacial SWCNT coverage (Fig. 4).In the case of the cell containing only supporting electrolytes (Fig. 4A, Scheme 1A and D), no signicant change in C int is seen as a function of increasing SWCNT concentration.A small ascendant trend is initially seen upon increasing the SWCNT density/thickness to 5.55 mg cm À2 in the completely reduced state (À0.1 V, black squares in Fig. 4A), though C int subsequently decreases upon further increasing the SWCNT density/thickness.However, when redox couples are present in both liquid phases (Fig. 4B, Scheme 1B and E) or only the aqueous ferri/ ferrocyanide redox couple is present (Fig. 4C, Scheme 1C and F), an increase of C int is observed with increasing SWCNT lm density/thickness, considering that a thicker SWCNT lm would yield a higher effective areal capacitance (analogous to a porous material).For the highest SWCNT coverages (11.1 mg cm À2 and 22.2 mg cm À2 ) there are exceptions to the trend of increasing C int in the reduced states: À0.10 V in Fig. 4B (both redox couples) and À0.10 V and +0.05 V in Fig. 4C (only the aqueous redox couple).The general trend, however, indicates an interaction between the electrochemical reduction of ferri/ ferrocyanide and the SWCNT lms, resulting in their charging.

In situ Raman spectro-electrochemical study at interfaceassembled SWCNT lms
In a conventional 3-electrode conguration, where SWCNTs are deposited on a solid electrode, the SWCNT Raman bands intensity dependence on electrochemical charging follows a bell-shape curve.Attenuation of the Raman intensity can be explained by lling/depleting of the Van Hove singularities. 40In situ Raman spectro-electrochemistry can be applied to the investigation of interfacial ET reactions between DMFc (oxidation to DMFc + ) in DCB and Fe(CN) 6 3À (reduction to Fe(CN) 6 4À ) in the aqueous phase, by monitoring the evolution of the Raman signals during potentiodynamic measurements in realtime, with a resolution of a few seconds. 45For this reason, interfacial ET in the presence of a SWCNT lm was also investigated using in situ Raman spectroscopy.To the best of our knowledge, the use of in situ Raman spectro-electrochemistry to probe ITIES assembled SWCNTs has not been reported previously.
The spectro-electrochemical cell composition was: Ag/ AgCl (s) |0.1 M K 3 (Fe(CN) 6 ) + 10 mM K 4 (Fe(CN) 6 ) + 0.1 M LiCl (aq) |5 mM BTPPATPFB + 10 mM DMFc + 2 wt% PVC (DCB) |1 mM BTPPACl + 10 mM LiCl (aq) |Ag/AgCl (s) .Fig. 5 shows the corresponding time-resolved Raman spectra at different potentials during the forward scan.As can be observed, the Raman intensity of all the characteristic SWCNT bands: radial breathing mode (RBM) at 155 cm À1 and 175 cm À1 , disorderedinduced feature (D band) at 1345 cm À1 , tangential displacement mode (G À and G + bands) at 1570 cm À1 and 1592 cm À1 and Dband overtone (2D band) at 2675 cm À1 decreases as the Galvani potential increases.The changing intensity of the Raman bands is expected to display reversible behaviour.A monotonic blue shiing of the G + band would indicate that SWCNT lm was pdoped. 43,44As can be seen in Fig. 5, the G + band was not shied during the electrochemical process.Therefore, it can be concluded that the SWCNT lm assembled at the DCB/water interface is n-doped during the interfacial ET reaction.Fig. 6 shows a reversible CV of DMFc oxidation to DMFc + and subsequent re-reduction, with the intensity of the tangential displacement mode, the G + band (1592 cm À1 ), at each Galvani potential, overlayered.
During the oxidation of DMFc, from À0.25 V to +0.  is avoided as the Fe(CN) 6 3À : Fe(CN) 6 4À ratio is 10 : 1.The in situ Raman data indicates that a net n-doping of the SWCNT lm occurs in the presence of both redox couples (aqueous and organic), whereas p-doping is suggested by the EIS data reported in the case of the aqueous redox couple alone, see Fig. 4.This implies that a spontaneous positive charging of the SWCNT lm is induced by contact with the aqueous phase redox couple, but a net negative charging of the lm occurs when the stronger, organic phase electron donor is present.This mechanism is consistent with the equilibration of Fermi levels recently discussed for adsorbed Au nanoparticles at the ITIES. 58

Conclusions
Aqueous and organic model redox species were added to both solution phases to characterise electrochemically active SWCNT lms, of varying thicknesses, at the DCB/water interface.Charge transfer through the carbon nanotube lms of differing thickness was studied with varying cell compositions and a doping/charging interaction was found between the redox couples and the interfacial SWCNTs using cyclic voltammetry analysis.This doping/charging was indicated by the charge transfer interaction between the electrochemically controlled ferri/ ferrocyanide redox reaction and the SWCNT lms, causing its doping.A novel in situ Raman spectro-electrochemical technique was exploited to investigate interfacial ET at carbon nanomaterial-decorated ITIES.The in situ Raman results also support the hypothesis of electrochemical doping of the interfacial SWCNT lms due to the charge transfer.This doping indicates that the adsorbed SWCNT lms can act as a pseudocapacitor, showing a high area-normalised capacitance, in a novel liquid interface conguration.Moreover, a deeper understanding of how interface-assembled nanomaterials can enhance heterogeneous electron transfer at the ITIES will aid in developing suitable materials to act as interfacial catalysts for biphasic reactions between immiscible aqueous and organic phases.

Fig. 1
Fig. 1 CVs of interfacial oxidation of organic DMFc by aqueous ferricyanide in the absence (trace b) and presence of varying densities of interfacial SWCNTs (c-g).A CV with only the supporting electrolytes present and in the absence of interfacial SWCNTs (trace a) is shown for comparison.Scan rate ¼ 50 mV s À1 .The coverages/densities of SWCNTs at the DCB/water interface were 1.38 mg cm À2 (trace c), 2.77 mg cm À2 (trace d), 5.55 mg cm À2 (trace e), 11.1 mg cm À2 (trace f), and 22.2 mg cm À2 (trace g).

Fig. 2
Fig. 2 CVs of the interfacial reduction of Fe(CN) 6 3À to Fe(CN) 6 4À (A, Scheme 1C) and only the supporting electrolytes of 0.1 M LiCl and 10 mM BTPPATPBCl in the organic phases, respectively present in each phase (B, Scheme 1A).Varying densities of interfacial SWCNTs (b-f) were employed.Curve a shows the respective systems in the absence of interfacial SWCNTs.The SWCNT densities at the DCB/water interface were 1.38 mg cm À2 (trace b), 2.77 mg cm À2 (trace c), 5.55 mg cm À2 (trace d), 11.1 mg cm À2 (trace e), and 22.2 mg cm À2 (trace f).Scan rate ¼ 50 mV s À1 .

Fig. 3
Fig.3Nyquist plots determined from EIS measurements of the 5.55 mg cm À2 SWCNT film at the DCB/water interface, with both aqueous and organic redox couples present (Scheme 1B).Measurements were performed at +0.15 V and +0.05 V, corresponding to oxidation of DMFc and re-reduction of DMFc + , respectively.The equivalent circuit used to determine interfacial capacitance values is shown.
4 V, Fe(CN) 6 3À reduction occurs concurrently and the intensity of the Raman bands is attenuated.When the DMFc + is reduced back to DMFc, the G + band intensity also increases again.This charge transfer occurs over a potential range where the SWCNTs are neither functionalized nor destroyed, and therefore, the Raman intensity of the bands is reversible.The change in Raman intensity is more reversible during the n-doping of the lms in aqueous solution at solid/liquid interfaces, anodic polarization in aqueous solution can produce defects by oxidation of the SWCNT, or even lm damage when it is illuminated by a laser light. 44It is noteworthy, that generation of Prussian blue (iron(II,III) hexacyanoferrate(II,III)) is observed when the Fe(CN) 6 3À concentration is equal, or in a slight excess, over Fe(CN) 6 4À .In the present case the Prussian blue formation

Fig. 4
Fig.4Interfacial capacitance (C int ) values as a function of the SWCNT density/thickness at the DCB/water interface (1.38 mg cm À2 , 2.77 mg cm À2 , 5.55 mg cm À2 , 11.1 mg cm À2 , and 22.2 mg cm À2 ).The different symbols correspond to the different Galvani potentials, at which the capacitance measurements were performed.The capacitance was measured for different ITIES configurations: only supporting electrolytes (A), both aqueous and organic redox couples (B) and only the aqueous redox couple (C).

Fig. 6
Fig. 6 Evolution of the G + band (tangential mode, 1592 cm À1 ) intensity in the Raman spectrum of SWCNTs at the water/gellified DCB interface during potential cycling (black line).CV peaks correspond to interfacial ET between Fe(CN) 6 3À /Fe(CN) 6 4À and DMFc/DMFc + redox couples in the aqueous and DCB phases, respectively.The Raman intensity was normalized to the most intense peak.The CV was recorded at 5 mV s À1 starting from D w o F ¼ À0.25 V (blue line).